Dear Fernando,

The first option is to use all the pixels in both of your study areas.

The second is to use sampling points.

In the end, you can use Independent T-test (If your data are normally distributed. If they are not, so try to use Kruskal-wallis) to assess if they are different based on mean (you can do this with both of the above mentioned options)

Hello Mohammad,

Thanks for your answer but I am not sure if it will help me, maybe it is my mistake not to specified what I am trying to do.

As I mention I have a map of vegetation index trend values, I want to make an spatial analysis (spatial patterns) over 2 separated areas in the map, both areas are similar in characteristics (comparable) but different in size. So, I though that there should exist a statistical methodology to make that spatial analysis without a much bigger representation of the larger area. Because right now it is much easier to make conclusion from what I saw in the bigger area.

Hope I could made myself more clear.

Dear Fernando,

I appreciate it. What you are saying about "without a much bigger representation of the larger area" is not make sense to me...

There are numerous spatial analysis techniques that you can take into account. But as I said, what you are saying is none sense to me.

Dear Fernando,

I had similar questions in the past.

I suggest you extract the raster of only the areas you are interested in using a command similar to "Extract by mask" in ArcGIS.

Once you have the two rasters, you import them into R for statistical analysis. I usually import the raster into R and convert them to a dataframe.

In R you can first create graphs of the two series of data to appreciate the difference graphically. Then you can perform statistical tests (with different levels of significance). To choose the appropriate test (in this case you have two samples with different sizes) I suggest you look at this guide: http://www.biostathandbook.com/index.html

Hello Eugenio Straffelini , actually I always use R for my analysis and for now I was thinking on use simple descriptive statistics converting the raster in a data frame. I was wondering if I could apply a more advance method for the analysis but I am not strong in statistics. I will take a look on the text you just sent me.

Thank you very much.